Abstract Background Patients are increasingly taking an active role in the design and delivery of surgical research. Public communication of results should also be encouraged, but this is often limited to non-expert commentary. This study assessed the role of plain English abstracts disseminated via social media in engaging patients and clinicians in the communication of surgical research. Methods A three-arm randomized controlled trial with crossover of two intervention arms was performed. Manuscripts accepted for publication in BJS were allocated to one of three arms and disseminated via Twitter: plain English abstracts, visual abstracts and standard tweets. The primary outcome was online engagement (a composite of tweets, replies and likes) by members of the public within 14 days. The secondary outcome was online engagement by healthcare professionals. Results Forty-one manuscripts were randomized to plain English abstracts (14), visual abstracts (14) and standard tweets (13). The number of public engagements was low, with a mean of 1·8 (range 0–8), 2·5 (0–11), and 1·2 (0–4) for plain English abstracts, visual abstracts and standard tweets respectively. The mean number of engagements by healthcare professionals was 29·4 (6–66), 45·3 (6–161) and 28·8 (10–52) respectively. Overall, visual abstracts attracted a significantly greater number of engagements than plain English ones (P < 0·001). Conclusion Online, public engagement with surgical research was low. Overall engagement (predominantly from healthcare professionals) was enhanced by the use of visual abstracts.